There is a quiet asymmetry in how policy intelligence is done. Most teams start with the press — a Bloomberg alert, a local newspaper, a Twitter thread — and then reverse-engineer the underlying document. It works well enough for major reforms. It fails silently on the second-order changes that actually move portfolios.
The alternative is gazette-first. Start from the primary sources that carry legal force: official gazettes, regulator circulars, central bank bulletins, parliamentary orders of business. Contextualise with press only after the primary source has been read. In our own data, the lag between an African gazette publication and its first substantive mainstream press write-up is usually measured in days, and often in weeks. Teams working press-first are systematically late.
There is a second, more durable benefit. Primary-source monitoring compounds: every gazette entry read this month makes next month's classification model sharper, and the longitudinal record becomes something no press-first workflow can reconstruct after the fact. That record is the real product.